---

Introduction

    Cancer incidence and mortality increase dramatically with increasing age. In addition, cancer incidence and mortality rates in the elderly have increased considerably over the past four decades. Interest has developed in the evaluation of the economic outcomes of cancer care in addition to traditional clinical outcomes. Both intermediate and ultimate outcome measures of clinical and economic interest are available. Survival and life expectancy, along with quality-adjusted life years, represent useful measures of clinical outcome. Health care costs should consider direct, indirect, and intangible costs. Direct medical expenditures include both institutional and professional costs as well as the costs of drugs and home care. Decision analytic models represent one of the most valuable types of economic analyses permitting simultaneous assessment of both clinical and economic outcomes with or without quality-of-life adjustment. Cost-effectiveness analysis utilizes both survival and cost measures, while cost-utility analysis considers quality-adjusted survival and cost. Such models permit evaluation of the tradeoffs between clinical benefit and harm or cost among elderly patients with cancer. Further studies of the cost-effectiveness of cancer care in the elderly should lead to an increase in our understanding of the effectiveness as well as the costs associated with the management of cancer in the elderly. An examination of the scope of clinical and economic outcome measures pertinent to patients with cancer is worthwhile for a number of reasons. First, physicians may not appreciate some outcomes that are relevant to both patient care and health policy. Second, there is increasing emphasis on more objective measures related to clinical and economic outcomes. Finally, these outcome measures illustrate techniques in economic assessment including the powerful method of decision analysis that may assist both physicians and admin-istrators in providing optimal patient care.

Cancer in the Elderly

Cancer Rates

    More than 550,000 Americans will die of invasive cancer in 1998, of which approximately two thirds will be individuals 65 years of age and over.¹ Cancer incidence and mortality rates increase dramatically with age and increase more rapidly in men than in women.² Over the past four decades, age-adjusted cancer mortality rates have increased by 10% in the United States from 158 per 100,000/ year in 1950 to 174 per 100,000/ year in 1990.² During this time period, cancer incidence increased by 26% and cancer mortality by 15% in those age 65 years and over compared with a 10% increase in cancer incidence and a 5% decrease in mortality among those less than 65 years of age.² The cumulative probability of developing cancer from birth to death based on current cancer incidence rates is 46.6% in men and 38.0% in women.¹ Most of the cumulative risk occurs between 60 and 80 years of age where the probability of developing cancer is 36.4% in men and 22.5% in women.

Cancer Types

    The leading types of cancer associated with mortality in the elderly are those of the lung, colon, prostate, and breast.³ The greatest increase in mortality rates over the past four decades among those age 65 years and older has been observed for lung cancer in women (200%), central nervous system malignancies (67%), malignant mel-anoma (65%), and non-Hodgkin’s lymphoma (49%). Lesser increases in mortality in this age group have been observed for multiple myeloma (37%), lung cancer in men (34%), esophageal cancer (25%), renal cell carcinoma (25%), and prostate cancer (21%).² Age-specific cancer mortality rates for men age 65 and above are approximately double the corresponding rates for women.

The Cost of Cancer Care

Total Health Care Expenditures

    Data available from the US Health Care Financing Administration show that total health care expenditures in the United States now exceed $1 trillion annually, representing a tenfold increase since the 1970s.⁴ Current expenditures are equivalent to 14% of the gross domestic product, and one half of these health care expenditures are for individuals age 65 and over.

Cancer Care Expenditures

    Approximately 10% of health care expenditures in the United States -- approximately $100 billion annually -- is spent on cancer care. More than 90% of direct medical costs for cancer care is associated with the following cancer types: breast (24%), colorectal (24%), lung (18%), prostate (17%), and bladder (8%).⁴ Table 1 compares US direct health care expenditures for cancer care to that for all health care. Nearly two thirds of direct health care expenditures for cancer care is for hospital care, while one fourth of expenditures is for physician services. The remaining 10% includes the cost of drugs and home health care.

Table 1. -- US Direct Health Care Expenditures
	Cancer	All Health Care
Hospital care	65.2%	49.0%
Physician services	24.1%	24.5%
Drugs	3.9%	10.5%
Nursing home care	4.9%	10.1%
Other professional services	1.9%	5.8%
Source: National Center for Health Statistics, 1990.

Distribution of Cancer Costs

    The distribution of health care costs for cancer care is not uniform over the natural history of the disease.^5,6 Cancer care costs are greatest during the first six months following diagnosis at the time of disease staging, primary treatment, and adjunctive therapy. The next greatest period of cost is during the six months prior to death in those who develop recurrent disease. Paradoxically, the greatest total cost is associated with the diagnosis of early-stage disease due to the long survival and lengthy course of follow-up. Alternatively, the greatest average annual cost is seen in patients diagnosed with advanced disease who often receive continuous and varied palliative treatments.

Health Care Outcome Measures

Clinical Outcomes

    Health care outcomes of importance include both clinical and economic measures. Effectiveness is the measurement of the outcome of cancer treatment in the population. This must be distinguished from efficacy, which represents the outcome apparent in a sample of the population in the framework of a clinical trial. As shown in Table 2, health care outcomes may be assessed utilizing both intermediate or ultimate outcomes as well as those that combine clinical and economic measures. Intermediate outcome measures are of value because of their rapid measurement and ready availability as well as their general association with the ultimate outcome of interest. Unfortunately, such intermediate measures do not always predict ultimate outcome, therefore limiting their value. Clinical outcomes may be expressed in terms of either quantity or quality.

Table 2. -- Cancer Outcome Measures
Outcomes	Clinical Measures:		Economic Measures
	Quantity	Quality
Intermediate	Response	Toxicity	Charges
Ultimate	Survival	Quality of life	Direct costs
	Life expectancy	QALY*	Indirect costs
Combined	Cost-effectiveness	Cost-utility	Cost-benefit
*Quality-adjusted life years.
From Lyman and Kuderer.21 Reproduced with permission.

    Quantitative Outcomes: Quantitative clinical outcome measures include intermediate measures such as objective tumor response, duration of response, or the time to progression of disease. The most commonly utilized meas-ure of ultimate clinical outcome is survival. Survival may be measured as overall survival, relative survival (where other causes of death are ignored), or disease-free survival (where both death and recurrence are considered adverse events). An alternative ultimate measure of clinical outcome is life expectancy or the average number of years of life remaining at a given age. The life expectancy from birth for both men and women in the US population is shown in Fig 1. Currently, a 65-year-old individual has an average of between 15 and 19 years of life remaining (Table 3). As can be seen, life expectancy from birth has increased progressively since the turn of the century.

Table 3. -- Average Additional Life-Years From Age 65
Year of Birth	Men	Women
1900	11	12
1920	12	13
1940	12	14
1970	13	17
1990	15	19

    Under conditions associated with a constant mortality rate over time, survival can be approximated by an exponential survival function. The requirement for a relatively constant mortality rate is often satisfied in patients with serious chronic diseases such as advanced cancer. When satisfied, the exponential survival function offers several attractive features for outcome measurement. The overall mortality rate is the sum of the age-specific mortality rate and the disease-specific mortality rates. The latter can be considered as the sum of the mortality rates associated with a specific type and stage of cancer and the rates associated with any comorbid diseases (heart disease, diabetes, etc). The life expectancy of a population whose survival is described by a declining exponential function is the reciprocal of the mortality rate.⁷ Such a process is therefore referred to as the declining exponential approximation of life expectancy (DEALE).

    Qualitative Outcomes: There is increasing interest in assessing the quality of clinical outcomes and adjusting the duration of estimated survival for the perceived quality of that time. Quality of life is a subjective concept, however, and its measurement is associated with both technical and conceptual difficulties. The factors that contribute to quality of life obviously constitute multiple dimensions or aspects of human existence. Crude measures of disease symptoms and treatment-related toxicity have been utilized by clinicians in assessing outcome from clinical trials. More sophisticated and valid measures of quality of life have been developed based on both psychosocial theory and decision theory. Multidimensional measures in the form of general as well as cancer-specific scales have been developed in an effort to assess the dimensions of quality of life. Table 4 provides a listing of the major quality-of-life dimensions described by Cella and his colleagues.⁸

Table 4. -- Major Quality of Life Dimensions*

Physical concerns (symptoms)

Functional ability

Family well-being

Emotional well-being

Treatment satisfaction

Sexuality/intimacy

Social functioning

*A global evaluation of QOL or total score is also given. From Cella.8

    Due to the difficulty in achieving agreement concerning the components of quality-of-life measures and their respective importance, more global measures of patient preference for a given clinical outcome may be sought.^9,10 Patient preferences are commonly measured as utilities along a scale from 0 (death) to 1 (full health). Such utilities permit an adjustment of the time spent in a given health outcome state. One tool used to measure patient preferences is willingness to pay to avoid or achieve a certain outcome. Willingness to pay has serious limitations as an outcome measure in that it not only requires the translation of clinical outcomes into monetary units, but also depends on an individual’s ability to pay. Various time trade-off methods have also been studied. The time in years of full health that is considered by the patient to be equivalent to the actual time in the diseased state is referred to as quality-adjusted life years (QALY). Quality of life may change over time depending on the basis of the disease state and accompanying treatment. Gelber et al¹¹ have utilized a measure termed quality-adjusted time without symptoms of disease or toxicity of treatment (Q-TWIST). Q-TWIST sums the product of the utility of a given state and the time spent in it over all health states (Fig 2).

Economic Outcomes

    Economic outcomes can also be expressed as intermediate outcomes such as charges or generated revenues. The ultimate economic outcome of interest is generally that of cost. As important as costs are to evaluating health care outcome, they are multifactorial and often difficult to measure.^12,13 Appropriate cost analysis depends on the perspective of the evaluation that can vary greatly from the viewpoint of the patient and family, the hospital, the physician, the third-party payor, or society as a whole. Cost-to-charge ratios based on reference data and measured charges are often used to estimate cost. Although such ratios may be relatively constant in certain settings, charges often have an inconsistent relationship to the costs of care. Therefore, the measurement of actual costs remains the goal of most economic analyses.

    Health Care Costs: Health care costs can be divided into direct, indirect, and intangible costs. Direct health care costs include both medical and nonmedical expenditures. Direct medical costs represent the costs of providing medical services for the prevention, diagnosis, treatment, follow-up, rehabilitation, and palliation of disease. Direct nonmedical costs represent additional expenditures incurred while receiving medical care such as transportation costs, daycare, etc. Indirect health care costs include those associated with the morbidity of illness and treatment and the economic impact of death from disease or treatment. Morbidity costs include the economic value of days of work lost due to illness. Mortality costs include the value of the economic output lost because of premature death. Such indirect costs are difficult to measure and require various assumptions about future economic productivity. Even more difficult to measure are the intangible costs associated with pain and suffering and loss of companionship. Meas-uring such costs again represents efforts to assess the impact of the disease or its treatment on the patient’s quality of life. It is often difficult to express such concerns in monetary terms. Due to the difficulty in assessing indirect and intangible costs, most economic analyses focus on the measurement of direct medical expenditures.

    Direct Medical Costs: Direct medical costs represent the most accessible of economic measures and are generally divided into institutional, professional, drug, and home health care costs.¹⁴ In an effort to measure the total operational costs of an institution, both direct and indirect institutional expenses may be considered. Direct institutional expenses are those related to the direct care of a patient and for which a charge may be generated, such as the nursing care unit, pharmacy, blood bank, radiology, or laboratory.¹⁵ Indirect institutional expenses are those costs that only indirectly impact on patient care and for which a charge is not generated, such as administration, engineering, and housekeeping. A major challenge in estimating operational costs of illness involves the process of allocation of the indirect expenses of the nonrevenue-generating support services to the revenue-generating service units in proportion to the expenditure of time and the utilization of resources. An accurate estimation of institutional operating expenses devoted to cancer care requires allocation of such indirect costs to revenue-generating services and ultimately to the level of specific diagnoses and procedures.

    Health Care Outcomes in the Elderly: Some specific issues must be considered when measuring health care outcomes in the elderly. Management decisions in the elderly must consider the greater age-specific mortality rates resulting in more limited life expectancy with increasing age. The limited treatment responsiveness of many malignancies affecting the elderly must also be considered. In assessing quality-of-life considerations, differences in drug metabolism and organ tolerance may increase treatment-related toxicity. Elderly patients are also more likely to have comorbid conditions that can decrease quality of life, increase toxicity, and further reduce life expectancy. While economic measures are fundamentally the same in older and younger patients, the greater potential for comorbidity as well as the limited resources and dependence on fixed incomes among the elderly should always be kept in mind.

Economic Analysis

    Economic analyses must consider both the clinical outcome and the economic outcome of interest.¹⁶ Economic analyses are useful when a management strategy is associated with the same or better outcome but at a higher cost or when it is associated with a lower cost but a worse outcome. In the first situation, the most efficient strategy will be the one with the lowest cost per unit of benefit, while in the latter approach, the most efficient strategy will be the one with the greatest benefit per unit cost. Types of economic analyses include administrative data sets based on the type of health care payor and retrospective studies of specific populations. There is also increasing interest in retrospective or prospective economic analyses in association with controlled clinical trials. Finally, economic analyses may involve the incorporation of available data into decision analytic models.

Clinical Decision Models

    Decision analytic models are valuable methods of economic analyses.¹⁷ Clinical decision models require explicit specification of the clinical problem or question in the form of a decision tree where each branching point represents a decision or chance event and the leaves represent endpoints or outcomes. Decision models must also specify the outcome probabilities and the values of those outcomes.¹⁸ Decision models can be analyzed in a variety of ways. Perhaps the most straightforward approach is the estimation of the expected value of each decision choice by the process of folding back. This involves multiplying the assigned outcome values of each branch by the probability of that outcome and then summation over all branches of the immediately preceding chance event. The sum represents the expected value of that branch, which now represents the outcome value, and the process continues. When a decision point is reached, the approach associated with the greatest expected value or the lowest expected cost is the preferred choice.

    Sensitivity Analysis: Perhaps the greatest strength of decision models of clinical problems is the ability to vary the assumptions related to model structure, probabilities or outcome values over a range of reasonable possibilities in a process termed sensitivity analysis. Such analyses also allow one to estimate the threshold at which the expected value of the decision choices are exactly the same. Decision analysis is particularly suited to economic evaluation by allowing simultaneous consideration of both clinical and economic outcomes in the form of a cost-effectiveness analysis. Decision analytic models also permit the incorporation of quality-of-life considerations or utilities in the form of a cost-utility analysis.

    Combined Outcome Measures: Cost-effectiveness and cost-utility analyses combine clinical and economic outcomes on the basis of cost and effectiveness or utility.¹⁹ Cost-effectiveness analysis measures the added clinical benefit (marginal benefit) and the added cost (marginal cost) of one strategy over the other. These two measures are then combined into a summary measure that can be based on either cost or effectiveness. Cost-based analyses compare management strategies on the basis of the cost for each unit of benefit, eg, cost in dollars per year of life gained (marginal cost-effectiveness). Effectiveness-based analyses compare management strategies on the basis of effectiveness for each unit cost, eg, years of life gained per dollar. In cost-utility analysis, quality of life is considered by assigning a utility value to each outcome state or estimation of quality-adjusted life years (QALYs). By measuring the added quality-adjusted benefit (marginal benefit) and the added cost (marginal cost) of an intervention, the cost per unit of quality-adjusted clinical benefit can be estimated (marginal cost-utility). Where appropriate, these values may be summed after weighting by the time spent in each outcome state. It must be noted that the summary measures of cost-effectiveness and cost-utility analyses are marginal outcome measures. Such marginal measures represent the incremental change in benefit with each unit change in cost, but they do not reflect the absolute benefit or cost that should also be assessed in any meaningful economic analysis.²⁰ A strategy may appear superior in terms of cost-effectiveness or cost-utility and yet have substantially lower absolute effectiveness or utility. It is important, therefore, to assess absolute as well as marginal measures of benefit and cost in such analyses.

    Cost Discounting: It may be necessary or appropriate to adjust changes in the cost or benefit measures for changes that occur over time. Cost discounting generally involves the adjustment of costs for the common preference of delaying present costs to the future. Future benefits may also be adjusted to the present based on the usual preference for immediate benefit.²¹

Decision Analyses in the Elderly

    Clinical decision analyses are of particular use in evaluating management strategies in the elderly. Such decisions involve the consideration of several issues of importance to the elderly patient with cancer. Cost-effectiveness analysis permits an evaluation of the trade-off between what is best for the patient (such as greater life expectancy or quality of life) and what is best for society (such as lower cost). For an individual patient, such analyses permit both an evaluation of what is most effective and what is least harmful as well as a distinction between the harmful effects of the disease and the toxicity of treatment.²²

    Several factors of importance to the elderly cancer patient must be considered in any cost-effectiveness analysis. The greater prevalence of cancer, which increases rapidly with age, yields a higher predictive value for any positive screening or diagnostic test. The types of cancer that affect the elderly are often ones that benefit the most from early diagnosis and treatment rather than those that present as advanced disease. Such studies must also consider the limited life expectancy of the patient and any comorbid conditions.²³

Conclusions

    Cancer care is associated with both clinical and economic outcomes of interest. There is increasing interest in measuring the impact of cancer and its treatment on the quality as well as the quantity of survival. Methods are available to evaluate management strategies based on both clinical and economic outcomes. These methods have only recently been applied to the study of cancer care in the elderly. With further use of such methods, we can anticipate a substantial increase in our understanding of the effectiveness and costs of managing cancer in the elderly. This knowledge should aid clinicians and health care planners in providing optimal quality and cost-effective care to the elderly patient with cancer.

Appreciation is expressed to Dorothy Allen for her excellent technical assistance in the preparation of the manuscript. This paper was presented in part at the Third International Conference on Geriatric Oncology, Tampa, Florida, USA, November 1996.

References

1. Landis SH, Murray T, Bolden S, et al. Cancer statistics, 1998. CA Cancer J Clin. 1998;48:6-29.

2. Miller BA, Ries LAG, Hankey BF, et al, eds. SEER Cancer Statistics Review: 1973-1990. Bethesda, Md: National Cancer Institute, NIH Pub No 93-2789, 1993.

3. Balducci L, Lyman GH. Cancer in the elderly. Epidemiologic and clinical implications. Clin Geriatr Med. 1997;13:1-14.

4. Brown ML, Fintor L. The economic burden of cancer. Cancer Prevention and Control. New York, NY: Marcel Dekker; 1995.

5. Riley GF, Potosky AL, Lubitz JD, et al. Medicare payments from diagnosis to death for elderly cancer patients by stage at diagnosis. Med Care. 1995;33:828-841.

6. Taplin SH, Barlow W, Urban N, et al. Stage, age, comorbidity and direct costs of colon, prostate, and breast cancer care. J Natl Cancer Inst. 1995;87:417-426.

7. Beck JR, Kassirer JP, Pauker SG. A convenient approximation of life expectancy (the "DEALE"). I. Validation of the method. Am J Med. 1982;73:883-888.

8. Cella DF, Bonomi AE. Measuring quality of life: 1995 update. Oncology (Huntingt). 1995;9(suppl 11):47-60.

9. Read JL, Quinn RJ, Berwick DM, et al. Preferences for health outcomes. Comparison of assessment methods. Med Decis Making. 1984;4:315-329.

10. Weeks J. Measurement of utilities and quality-adjusted survival. Oncology (Huntingt). 1995;9:67-70.

11. Gelber RD, Goldhirsch A, Cavalli F. Quality of life-adjusted evaluation of adjuvant therapies for operable breast cancer. The International Breast Cancer Study. Ann Intern Med. 1991;114:621-628.

12. Brown ML. The national economic burden of cancer: an update. J Natl Cancer Inst. 1990;82:1811-1814.

13. Schuette HL, Tucker TC, Brown ML, et al. The costs of cancer care in the United States: implications for action. Oncology (Huntingt). 1995;9(suppl 11):19-22.

14. Vincenzino JV. Health care costs: market forces and reform. Oncology (Huntingt). 1995;9:367-368.

15. Lyman GH, Kuderer NM, Greene J, et al. The economics of febrile neutropenia: implications for the colony-stimulating factors. Eur J Cancer. 1998. In press.

16. Schulman KA, Yabroff KR. Measuring the cost-effectiveness of cancer care. Oncology (Huntingt). 1995;9:523-530.

17. Lyman GH. Essentials of clinical decision analysis: a new way to think about cancer and age. In: Balducci L, Lyman GH, Erschler W, eds. Comprehensive Geriatric Oncology. London: Harwood Academic Publishers; 1998:7-17.

18. Pauker SG, Kassirer JP. Decision analysis. N Engl J Med. 1987;316:250-258.

19. Detsky AS, Naglie IG. A clinician’s guide to cost-effectiveness analysis. Ann Intern Med. 1990;113:147-154.

20. Udvarhelyi IS, Colditz GA, Rai A, et al. Cost-effectiveness and cost-benefit analyses in the medical literature. Are the methods being used correctly? Ann Intern Med. 1992;116:238-244.

21. Lyman GH, Kuderer NM. The diagnosis and treatment of cancer in the elderly: cost-effectiveness considerations. In: Balducci L, Lyman GH, Erschler W, eds. Comprehensive Geriatric Oncology. London: Harwood Academic Publishers; 1998:533-543.

22. Russell LB, Gold MR, Siegel JE, et al. Role of cost-effectiveness analysis in health and medicine. Panel on Cost Effectiveness in Health and Medicine. JAMA. 1996;276: 1172-1177.

23. Extermann M, Overcash J, Lyman GH, et al. Comorbidity and functional status are independent in older cancer patients. J Clin Oncol. 1998;16:1582-1587.

Glossary

Cost: resources spent to purchase services or other resources including direct, indirect and intangible components.

Cost-to-charge ratio: method of estimating cost based on charges and assumed distribution of costs per unit charge.

DEALE: declining exponential approximation of life expectancy that estimates survival with the assumption of a constant mortality rate whose reciprocal is the life expectancy.

Decision analysis: form of analysis based on decision theory incorporating a formal decision structure (tree) and specified probabilities and outcome values to evaluate treatment alternatives.

Direct costs: medical and nonmedical costs associated with the provision of medical services for the prevention, diagnosis, treatment, follow-up, rehabilitation, and palliation of illness.

Direct institutional costs: costs associated with the direct provision of services to the patient and for which a charge may be generated.

Discount: adjustment in benefit or cost in the future relative to benefit and cost in the present.

Effectiveness: measurement of treatment effect in the population outside of a clinical trial.

Efficacy: measurement of treatment effect within a sample of the population in the framework of a clinical trial.

Expected value: calculated value at a chance or decision point in a decision model found by summing all products of outcome values and probabilities of branches distal to the point.

Indirect costs: cost associated with the morbidity or mortality of illness beyond the direct provision of care.

Indirect institutional costs: costs associated with the operation of the institution not directly associated with patient care and for which a charge is not generated.

Intangible costs: poorly defined costs associated with illness including pain and suffering and loss of companionship.

Life expectancy: average number of years of life remaining at a given age.

Marginal benefit: difference in benefit between two strategies.

Marginal cost: difference in cost between two strategies or treatments.

Marginal cost-effectiveness: difference in cost to achieve an additional amount of benefit with a treatment strategy usually expressed in dollars per year of life gained.

Marginal cost-utility: difference in cost to achieve an additional amount of quality-adjusted benefit with a treatment strategy usually expressed in dollars per quality-adjusted life year (QALY) gained.

Q-TWIST: quality-adjusted time without symptoms of disease or toxicity of treatment representing the sum over all time intervals of the product of time and utility.

Sensitivity analysis: process of assessing the change in expected value or threshold values based on variation of the probabilities or outcome values assumed in a decision model over a range of possible values.

Threshold: value of a variable evaluated in a sensitivity analysis where the expected value of the decision choices are exactly equal.

Time trade-off: method for assessing patient preferences by estimating the time in full health that is considered equivalent to actual time in the diseased state.

Utility: measured patient preference for a given health outcome state.